In 2013, we saw many CTP clients using OpenStack based tools for their private cloud initiatives. Fast forward to 2016 and public cloud is 3.5x the size of private cloud adoption, according to a Wikibon report. As more features and functions are released by the public cloud providers, enterprises are rightfully asking where the private cloud business case exists anymore.
Despite the increasing popularity of public cloud’s on-demand, pay-per-use model, private cloud still has a place in the enterprise, and its adoption continues to grow across platforms. There remains legitimate reasons to own and manage your own technology infrastructure, and private clouds bring some of the benefits of cloud principles to such data center deployments.
For Episode 100 of The Doppler podcast, David Linthicum sat down with private cloud proponent John Engates, and private cloud skeptic Bernard Golden to moderate a discussion around the future of private cloud and its business value in IT’s dynamic landscape.
Meet our discussion contributors
John Engates, CTO, Rackspace
John joined Rackspace in August 2000, managing data center operations and customer-service teams. Today, he is active in extending Rackspace’s fanatical support to clouds like AWS and Microsoft Azure and supporting OpenStack private cloud for customers all over the world. John is also an internationally recognized cloud computing expert and a sought-after speaker at technology conferences, including CA World, the Goldman Sachs Techtonics Conference and Cloud Expo. He speaks on the future of cloud computing, enterprise cloud adoption, data center efficiency, green data center best practices and more.
Bernard Golden, CEO, Navica
Named by Wired.com as one of the 10 most influential people in cloud computing, Bernard has extensive experience working with organizations to help them adopt and integrate cloud computing effectively. Bernard serves as VP of strategy for ActiveState Soft- ware and as CEO and founder of Navica. He is the author of four books on virtualization, cloud computing and AWS topics and a regular contributor to CIO magazine. His work has also been published by the Harvard Business Review and The New York Times.
David Linthicum: What is the business case for private cloud and where is it best applied?
Bernard Golden: The business case for private cloud is typically centered around an IT organization that has constraints on a particular set of solutions that preclude the use of a public cloud environment. It may be that they have compliance requirements, or they’re located in a geography that insists data not be located outside of the national boundary, and there’s no viable cloud provider within that national boundary. If they want to use cloud computing, they have to do something else.
Another common constraint is that organizations want to build applications that communicate with existing applications residing within their data center, and the latency for the network traffic would be too great to be able to place a new application outside the boundaries of their data center.
A final common constraint is the concept of data gravity. If you have a huge amount of data sitting in your data center, it is impractical to migrate that to a public cloud provider, so you need to write applications where that data storage exists. All these imposed constraints require an ideal solution to be executed within an organization’s own data center.
John Engates: I would agree with those constraints. I would also say there are other drivers that lead people to choose a private cloud over a public cloud. For example, companies that have large existing data centers with data center assets, such as a long term contract for a colo, or a half-empty company owned data center, don’t think it makes financial sense to spend money on public cloud when you already have sunk costs. Other times, companies are measured on EBITA and that will drive them to choose and prefer capital expenditures over operational expenditures. We see companies all the time in private equity situations where they are being measured by CapEx dollars, so in many cases it makes a lot more sense to own and buy the equipment than it does to rent it from a cloud provider.
Other cases involve situations where companies don’t want to risk the government having access to their data. This is something that is been on the minds of everyone since the revelations of Edward Snowden. A couple of years ago, people worried that there might be a secret subpoena that had been issued for access to their data and they had no way to know that their data was being accessed via their cloud provider. Security and compliance is another business case for private cloud. In certain cases, companies need to be able to demonstrate strict security or compliance requirements that various clouds may not have the capability to live up to today.
I think the bigger drive for private cloud is cost. We find companies telling us that at certain points in their cloud journey when they look at cost comparison or TCO, their cost reductions can be 20%, 30% or even 40% to 50% compared to public cloud. This may be due to the size of the company, or the stage of the company, but the point remains that many companies feel the pain of cost and think they can save money in the private cloud.
DL: Is the antithesis of private cloud moving to the public cloud, or is it leaving things on the legacy system? What are the options if we’re not moving to a private cloud?
BG: I think many companies are embracing public cloud computing; that is pretty clear from the kind of growth the public providers are seeing. For many companies that already have assets in place, their question is going to be, do I leave it as it is and accept the system I have, or do I make the extra investment to bring cloud capabilities to that infrastructure. My prediction is that more companies are going to look at the assets they have, and make a decision about either moving to the cloud, or look to find a SaaS alternative.
Typically, moving to the cloud requires a significant investment both in infrastructure and in skills. Running a cloud program is hard, and organizations will have to evaluate when they reach what I call the application life stage when an application comes to the end of a hardware cycle, or it is time for a mandatory upgrade. At that point, they will make a decision. My sense is that a lot of companies won’t do too much until they have to.
DL: John, Rackspace is placing bets in both public and private clouds. What do you see are the tradeoffs and what are you advising your clients in terms of where to put their stuff? Is private cloud really the anti-public cloud or is it anti-legacy?
JE: I think the cloud is an upgrade over your existing legacy. It certainly makes your legacy environment a lot more accessible to you and your users. It makes it something that is more agile and nimble and automatable, and those are all the good things that people love about public cloud. At Rackspace, we are seeing companies upgrade their on-premises environments to be a private cloud or more like a private cloud. Many companies are working hard to figure out how to make private, on-premises environments more cloud like.
I do agree that public cloud makes complete sense if you’re a small clean company or a startup without existing data centers. IT is messy. Sometimes existing IT was built over dozens of years, with multiple generations of hardware, software and networking gear. There are also politics involved. People have built their careers around a particular stack of technologies. This can lead companies to use both private and public cloud. Many companies that we encounter are using public cloud for some portion of their IT infrastructure and private cloud for other portions. They intend it to be a long term, multi-year cloud arrangement.
DL: Is security still a compelling argument for leveraging private cloud? If so, what specifically is the issue?
BG: A lot of what is called security is really compliance, which is a very important reason why companies might require systems to run on their own premises. I think of security as how resistant the resources are to attack, or how well they’re protected against intrusion. I think it is dif cult to make the claim that a private cloud can execute security better than what the public cloud providers do on that dimension. We haven’t seen anywhere near the kind of intrusions and hacking into public providers that we’ve seen in companies running their own data environments. From what I’ve seen, there seems to be a gradual acceptance towards public cloud.
JE: I agree with Bernard on some aspects. Public cloud providers are very good at security. Google, Amazon and Microsoft are top notch at running security at scale across very large environments. They have a very large investment in the security operations, as we similarly do at Rackspace. Having said that, there are some companies that have very specific security requirements. For example, financial services institutions are responsible for certain regulated data that they don’t want to hand over to a third party. In some cases it is just untenable for them to let that data out of their own facility.
I think that clouds can be secure, but public clouds can also be insecure, depending on how they’re operated. This is something we’ve found at Rackspace as we’ve interacted with customers that are using both public and private clouds. A lot of the security depends on what is hosted in the cloud and how that application has been built and architected. The cloud doesn’t magically make an application secure or insecure; it has a lot to do with how you manage it, the policies around it, and compliance sometimes is what forces your hand.
DL: Are we dealing with practical reality, or are we protecting our jobs and egos more than we are protecting data?
BG: For some companies there are a portion of their applications that must reside on premise for these compliance and security reasons we mentioned. The key question is, do those requirements apply across the entire application fleet? Warren Buffett says you should know to never ask a barber if you need a haircut. Similarly, if you ask the people who run the on-premise systems if you need to keep your stuff on-premise, the answer is probably going to be yes. If you’re a senior IT executive that is under a lot of pressure, one of the questions you ask is how do you manage costs, but also how do you help your organization become a digital enterprise. Unless you really drill down and look at what proportion of your application portfolio requires on-premise solutions and what proportion can be done in a different fashion, you just can’t accept the face value you get back.
JE: I think that is what companies are doing when they’re evaluating private cloud. They’re looking at what they have today and what proportion could go to public cloud. People are making public cloud the default for certain applications if it is customer facing or a mobile application. But at the same time, there is a lot of gravity around those existing applications that have run the business for a long time. As good as cloud is, and as good as a lot of the cloud providers and vendors are, there aren’t always perfect applications that can replace everything in your environment in the public cloud. There are a lot of things that companies continue to run on a mainframe or on some legacy platform that sits side by side with other applications in their environment. Those things are very difficult, if not impossible, to pick up and move to a public cloud.
Private cloud checks off a lot of the other boxes that organizations are looking for–in the form of automation and agile infrastructure, potentially lower costs and geographic requirements–because it is in their own data center. It is not just one of these things that pushes people to private cloud, it is the fact that it hits a lot of these things, and for certain core applications it is just the easiest route. The path of least resistance is sometimes the right path for companies.
DL: What is the cost and agility argument for private clouds? Why move from a legacy environment where you own the hardware and software, to another environment where you still own the hardware and software? What would be the internal argument to move into private clouds?
BG: To be honest, I think that is a big challenge. An investment is required to move from standard issue virtualization where system administration is doing manual tasks, to providing cloud capabilities. There isn’t a magic button on the side of the server that you can press to make it cloud capable. It requires software infrastructure, hardware infrastructure and you will certainly have to train up people’s operational skills. You need to make a strong nancial assessment. Ask yourself, will I get enough value out of that investment and create capabilities beyond what I already have in place? That seems like an ongoing challenge.
JE: One of Rackspace’s strong suits is the support that we provide for our customers. Lack of expertise is often the number one reason for not moving to cloud at this stage. Rackspace has actually built a model for delivering private cloud that we believe gives users a lot of the benefits of the private cloud, but with the turnkey nature of public cloud. We bring all the expertise you need, we do the automated upgrades and updates to the private cloud software. We make it as painless as possible for a company to deploy, and we can even bring in hardware, if a customer wants the hardware. They don’t even have to own it, they can rent it from Rackspace, whether it is in our data center or theirs.
I think there are models with OpenStack or private clouds in general, where agility comes from the fact that users look at that private cloud just like they would a public cloud. They see an endpoint on an API; they see the automation tools they’re using today– like Chef, Ansible, Salt and all the tools that they’re comfortable with–working equally well on public or private cloud. Again, the only difference is that it physically sits in your data center and it makes the life of certain people in the IT organization a whole lot easier, because they don’t have to make the case that everything has to change within the environment. Only certain things have to change, and that is just easier to swallow.
Whether or not you believe public cloud is the long term future, which I agree that long-long-long term it may be–10,15 or 20 years–certainly public cloud has a bigger and bigger role over time, but that doesn’t mean private cloud isn’t a very big and growing business. The folks at ITC say that private cloud is growing at a rate of 35%, reaching ~$40 billion by 2019. It is a very vibrant growing business and it just proves that there is a demand for it.
DL: Where do you see the future of cloud going, and how quickly do you see it moving from private to public?
BG: I think you’ve identified the key issue, which is there’s going to be a huge amount of computing going on in public cloud. If we look at different milestones in 10 to 15 years, the question then is what percent of the total application fleet will remain on premise? Will it be 80%, or 50% or even 20%? My estimation is that it is going to be lower than most people in the industry expect. Many vendors and analysts use a mental model that thinks some stuff will be in the public cloud, but 80% of computing will still be done on premise. I think that is vastly optimistic.
DL: John, can you expand on your growth projections? You are a bit more conservative than Bernard, but you do think public cloud will become kind of the default platform?
JE: I think the public cloud is growing nicely, but I was just making the case that the private cloud has continued to grow as well, and there are major companies adopting it every day. There is a statistic that was thrown out at the last OpenStack summit–I think it said 50% of the Fortune 100 are using OpenStack. Now imagine that same 50% are probably doing something with Amazon and probably doing something with Azure and possibly experimenting with Google and others. The point is that companies are using private cloud in conjunction with public cloud. Private cloud has a big role to play and it is being used for some very large scale workloads. There are big banks that are using private cloud, there are major telecom companies and there are other very large retailers who look at Amazon as a potential competitor and therefore gravitate to the private cloud. It is always surprising to me, just how many companies
are leaning into both private cloud and public cloud. It doesn’t feel to me like the private cloud is dying anytime soon.
DL: What advice would you give to enterprise IT looking to adopt the private cloud?
BG: I think it is important to do a really clear eyed analysis of your current application portfolio, and of the application portfolio you’re likely to develop. What proportion of those require cloud capabilities and have constraints that require them to remain on premise? Think about your total spend and portfolio scope to determine what investment or spend you can afford to make to support that, and what you would consider using the public cloud for. What I tell CIOs is, don’t ask the infrastructure group what they think, because they’ll have a natural inclination to come up with a very conservative approach.
JE: I would agree. I think you need to come in with your eyes open and understand the total picture of your IT requirements, where it is going and how much it is going to cost. Take inventory of your skills to understand what you’re good at and what you’re not good at. If you’re a smaller organization, by all means don’t go build a data center, go to the public cloud. But if you’re an organization that has substantial IT, I think a private cloud can play an important role for a long time to come in that environment.
The transcript has been edited to publication level with respect to capitalization, punctuation, grammar, spelling, and paragraphing.